Pump suction line full vacuum in transient simulation

Is the vacuum sufficiently strong to boil the water?

Whatever the condition, the negative pressure cannot be less than -1 bar, i.e., 0 bar absolute pressure in the described system. Most pipelines and components can take this. Is there something else invoved?

Bioramani

In general a vacuum in the suction line can occur when the pump starts and its rpm go up to fast. At shut down the incoming flow has an inertia and will continue to flow direction pump for a short time depending on inertia, pump resistance, check valve design and roughness of pipe. What dimensions has the suction line (L= Di=)? Where do you obtain by simulation a vacuum? At pipe entry? pump inlet? What kind of valve did you consider and how did you choose the model, was it made by you or was it proposed by the soft? Where did you place the check valve?

I make dynamic simulations since a long time and I always give the counsel "NOT to believe what the soft gives as result if qualitatively it does not seem to be logical".

Logical means, in your case,the respect of basic hydrodynamic principles.

The suction header is approx 100m long, diameter is 1500mm reducing to 800mm at branches in front of each of the 3 duty and 2 stand by pumps. The pressure drops all the way to -1.0 bar so the water will boil for sure. Pipe material is GRP and the simulation took into account roughness (Darcy-Weisbach e) of 0.03mm. Almost full lenght of this suction line is under the vacuum, however the most severe conditions are just in front of the pump itself. Check valve was modelled to close instantenuosly and is placed in discharge pipe downstream from the pump, and the pump/motor set has its inertia of 290kgm2.

Before going into more details, have you computed the velocity of the fluid in the different sections?

Suction of a pump: Flow=1.75m^3/s Area d=0.8m 0.5026 m^2

V= 1.75/0.5026=3.48 m/s!!!

Main duct: Flow= 1.75*3 (3 pumps active)= 5.25 m^3/s Area d= 1.5 m 1.767 m^2

V= 5.25/1.767=2.97 m/s!!!!

For a suction line you are very high which justify the pressure drop. In general the velocities in suction lines do not go much over 2m/s.

If the pressure is such that the water boils then you cannot guarantee the correct function.

If you want to have same velocity allover the smaller diameters should be about 866mm, but according to above values both diameters should be increased: main about 1750..1800 and the pump connections to 1000. If the pump has only 800 as entry the connection between 1000 and 800 should be only as long as necessary for a correct flow pattern.

When I mentioned "inertia" I thought the fluid column inertia which will generate, when the valve closed, a pressure peak due to accumulated kinetic energy (water hammer effect).

If Re>Re*= 1300*d/h the tube can be considered a rough since the boundary layer has same order of magnitude as the roughness. Re*= 1300*1/(3 e-5)= 3.37E7 and the Re= 2.97*1/(1E-6)=3 E6 < Re* roughness has thus less importance.

Assuming the fluid is water at 20°C kinematic viscosity is 1E-6 m^2/s

λ≈ 0.309/(log(Re/7))^2=0.02743 Δp=(λ*L/d)*ρ*v^2/2=8230 N/m^2

an other source gives

λ≈ 0.032+0.221/Re^0.237=0.0384 Δp=(λ*L/d)*ρ*v^2/2=11535 N/m^2

This is, as order of magnitude, the pressure loss in the main. Water saturation pressure at 20°C is ≈2340 Pa so that as long as pressure is> it will not boil.

According to the computations (which you should check since an error can always appear) the losses are less atmospheric pressure around 1E5Pa. If results are correct simulation is wrong.

In all manuals for simulations it is strongly recommended to make some simple computations with classical equations in order to validate orders of magnitude.

Hope it will help.

Please keep informed.

I made a mistake, the flow is not 1.75 but 1.05 cms, therefore the velocities are around 2 m/s. As for your advice to compute the problem manually, I'd do that if I knew how , however since the project is outta my hands now, I just wanted to ask if anyone knew any similar case from practice, i.e. is it possible to have such severe vacuum in suction pipe, because in general it is connected to the huge atmospheric reservoir which, in my opinion, could act as sort of surge tank. Anyhow thanks for the advices it means a lot.

Look at the wikipedia page concerning the equations you have every thing you need for a "hand calculation".

Have you verified mines?

Yeah, I'd say your calculation is pretty much correct, the software calculation is resulting in around 1m headloss in suction pipe. However, the initial pressure is very low - like I said before, the water level in suction reservoir is, lets say at the elevation of pump center line (I wanted to check the worst case), therefore the starting conditions are already around -0.something bar. Hence the pressure drops very low during transient event at early stage of simulation, and never gets recovered.

Ill check wiki, thanx for the tip.

Regards

I think maybe I am being a bit dumb here and not understanding.

You say that you have 1 m head loss in the suction line and that reservoir level is roughly at pump centreline, so you have 9 m positive suction head.

The transient that you talk of is when the pump trips, is that right? If you have modelled instantaneous shutting of the NRV in the discharge line of the pump, what then causes the vacuum? The water in the suction line only ever has a forward velocity according to your model.

After pump stoppage, water initially coming towards the pump suddenly has nowhere to go. As water packs into a finite space upstream of the pump, it generates a high pressure pulse that propagates upstream at local wave speed, away from the pump causing the local pressure drop. As the pipe is relatively short and smooth (since the pipe is being modeled as new), there is nothing to slow the oscillation of this pressure wave down. So, 9m of positive head (0.9 bar) easily drops down to -0.9bar than goes back up, than drops down again...